Dante.1508 Posted September 4, 2020 Share Posted September 4, 2020 I'm on NA servers and its just as bad there. Link to comment Share on other sites More sharing options...
Craine.8162 Posted September 4, 2020 Share Posted September 4, 2020 The last couple of months i barely played anything apart from fractals and raids and i never had any connection issues.yeterday i went to vabbi to do the forged with fire daily. and the lag spikes were ridiculous. 2-3s of lag every ~10s. it was just unplayable. Link to comment Share on other sites More sharing options...
Ayrilana.1396 Posted September 4, 2020 Share Posted September 4, 2020 @maddoctor.2738 said:@Ayrilana.1396 said:Because that’s not how it works. Servers generally do not pick and choose who has issues and who doesn’t. If there’s a series of input lag issues caused by a server, it for the most part affects everyone connected to it. The fact that some players consistently do not experience any issues when on the exact same map instance doing the exact same meta means that it’s unlikely to be a server issue. Are you sure that it's a fact players aren't experiencing any issues? Did you try a tracert or a ping plotter recently to compare it with one from last year? How exactly are you sure you aren't affected by server issues?My ping has remained the same for the past year. Link to comment Share on other sites More sharing options...
MattDu.7123 Posted September 6, 2020 Share Posted September 6, 2020 post removed Link to comment Share on other sites More sharing options...
Inculpatus cedo.9234 Posted September 6, 2020 Share Posted September 6, 2020 Well, the route to each individual map may be different. If a Tier 1 backbone is having issue, it will affect many players from different parts of the world, but not others who used a different Tier 1 backbone to get to the same destination. There are only a few Tier 1 backbones (5 or 6?), and if a few are having issue, it could cause a lot of grief. But, still some players would not be affected, as they used a different route.No one knows, though, for sure; not even ArenaNet, as they have stated several issues seem to be at play at once. Link to comment Share on other sites More sharing options...
Zok.4956 Posted September 8, 2020 Share Posted September 8, 2020 @Ayrilana.1396 said:@Maikimaik.1974 said:@Ayrilana.1396 said:@Maikimaik.1974 said:@Ayrilana.1396 said:@Maikimaik.1974 said:@Ayrilana.1396 said:@Trevor Boyer.6524 said:@Ayrilana.1396 said:@Trevor Boyer.6524 said:Yeah, I logged on today into the pvp mist area. My normal average ping of 70 on NA was bumping around near 500 avg. I've waited around logging in and out for hours. It's not my ISP. The problem is not on my side. This lag is unplayable.How do you know? What you said doesn’t prove that it couldn’t be your ISP. Just saw someone post in one of the couple dozen threads that they had an issue that tracked to the AWS servers, they contacted Anet, and it got troubleshooted to be with their ISP. As if everyone's home in 2020 didn't have multiple wireless connected devices on, all of the time, that weren't lagging at all.^ This works this way for years on end, nothing lags. Then suddenly at the very same time my Guild Wars 2 is getting 500 average ping, other players in Guild Wars 2 are also reporting that they are getting averages of 500 ping. And mysteriously enough, nothing else in my home that is connected to the wireless is lagging at all.Now in theory, 1+1=2, but I guess we would need to examine quite a bit further to make sure that this equation was true.I mean, I want to support the game and give the Arenanet team the benefit of a doubt as well, but come on now.Just because nothing else is lagging doesn’t mean that it’s not the ISP. The ISP is the one who routes your connection to the servers. Somewhere along that line could be having issues. There are also a lot more people using the internet which puts stress on the networks. This could exacerbate any issues. How hard can it be to understand that if dozens of people from all over the world experience severe lags at the exact same time, on the exact same map, it CANNOT be a player side issue. This has been happening for months, even before covid, it ONLY affects GW2 and only a certain set of maps.How can you seriously believe that this isn't a problem on Anets side? I'm speechless to be honest.I never said it was player side. You do realize that between the players and Anet is the ISP? How can it be the servers if it doesn’t affect everyone?How propable do you think it is that dozens of players from all over the world use the same ISP? Or that several ISPs lag at the exact same time?They all use the same infrastructure. How do you explain that this issue only affects Gw2? How do you explain that this issue only really affects a specific set of maps? (Anything PoF and newer).It doesn’t affect just GW2. If it’s just specific maps then it could very well be a PC issue. I’ve been on maps with people complaining about lag while I’m doing the exact same meta as them but experiencing no issues. Keep in mind that people have reported issues on older maps. This is obviously, without any doubt, a problem on Anets side. It cannot be an ISP issue or it wouldn't affect players from literally all over the world.Untrue. I don't understand why you can't just accept that there may be an issue on Anets side. What benefit do you get from stubbornly denying any evidence that this issue isn't on the players/IPSs side.What harm would be done by Anet looking into this issue?You are actively harming ANet by suggesting that they can't do anything about this. There's nothing more likely to turn people off a game than bad performance. Because that’s not how it works. Servers generally do not pick and choose who has issues and who doesn’t. If there’s a series of input lag issues caused by a server, it for the most part affects everyone connected to it. The fact that some players consistently do not experience any issues when on the exact same map instance doing the exact same meta means that it’s unlikely to be a server issue. There are from time to time software/systems bugs in GW2 that cause that one version of a map is in a buggy state, and other versions (different IP) of the same map are not and only players that are on the buggy map (IP) experience problems. The fact that some players do not think they experience problems does not prove anything about the problems that other players experience.Only players/accounts that are experiencing the problems can be used to make tests and to try to locate the problems a little bit more with testing.And we can not know from the outside (no admin/monitoring access to Anets systems) if it is a "server/instance issue" or a "network issue" or a "software issue". In fact, it could be anything and I am sure that there are players where their own bad Wifi/WLAN is causing the problems. But this also does not prove, that there are not also other problems on Anets side. So, until we can make tests, we can not rule out anything.I do experience often skill lag spikes (some several seconds long) in Thunderhead Peaks. So I made some tests during the time the lag spiked happend in my GW2 client and run a MTR to the map IP is was on. During the time the lag spikes happenend, the whole way the packets travel from my PC to the (first) IP of the Amazon network was clean. I had a small RTT (packet round-trip-time) of a few ms (milliseconds) without any spikes of long RTT and MTR had not one lost packet. During the whole time.I also made tests and saw that it is is somhow map specific. On some maps (i.e. core maps) I normally have no skill lag/lag spike, but in S4 maps (especially in Thunderhead Peaks) I had very often skill lag (spikes) of several seconds.Because we have several PC at home I also tested this with two PCs at the same time (one in core maps, one in Thunderhead Peaks, and I also switched the maps/PCs to make sure it is not PC/Hardware specific)): The PC that was in Thunderhead Peaks had skill lag (spikes), the other in core-maps had not.The MTR to the map IPs of the Thunderhead Peaks map and the core map travelled the same way through the internet to the same Amazon IP of the Amazon gateway.So, I am pretty sure, that, during my test, definitely not my internet-connection/access and definitely not "the Internet" (faulty or overloaded providers or provider interconnections) between my PC and the gateway of the Amazon network (Amazon cloud, or whatever you call it) caused the skill lag spikes. So, what else could it be?The only part the packets travel that I could not test further was inside the "Amazon Cloud/Network/Servers" where Anets game servers are.As a result, logic dictates it is has to be somewhere located in the "Amazon cloud" (network, servers, software) and as a result it is the responsibility of Anet to fix this.I only talk about my own tests. I can not prove/disprove if this is true for other players. I am sure there are players with bad WLAN/Wifi that is causing the same problems.Because that’s not how it works. Servers generally do not pick and choose who has issues and who doesn’t.Well, of course servers do this. There could be a lot of reasons, why only some users have issues/server problems and others don't. And the more complicated the server structure, the more complicated it could be to find the reasons and to fix the issues. The hardest issues to find are the ones that do not happen allways. And sometimes, the servers "pick and choose" in a nondeterministic way. Example: race conditions. https://en.wikipedia.org/wiki/Race_conditionWhat is different between PoF/S4 maps and other maps, that could be the reason? The map itself (how complicated it is to process it on server side and client side, how much data flows between client and server, the software code) and of course the IP of the map server (which probably means, that it is handled by a different map server/instance).It is not unusal (it has happened often in the past) in GW2, that some maps (map IPs) are stuck in some buggy state (caused by overload, software bugs etc) and other versions of the same map (other IPs) have no problem at all. As a result: Some players experience problems, others do not.I do not know the internal structure of the Anet game servers to make better educated guesses about the reason why this exact problems happen not to all players. It could be related to time-of-day or how much players are on the same map, or how much map-instances are running at the same time. Or something else. Whatever. But I do know that my own tests ruled out "The Internet" and my own equipment. So it has to be somwhere located in the Amazon network (Amazon/Anet) and it is the responsibility of Anet to fix this. Link to comment Share on other sites More sharing options...
Inculpatus cedo.9234 Posted September 8, 2020 Share Posted September 8, 2020 Amazon has a Tier 1 backbone; data travels through this, out of this, and then to the AWS servers.It is odd that some players in the same map (same IP address) experience issues while others do not.One would suppose that all players on a faulty map/server/IP address would all experience the issue. Link to comment Share on other sites More sharing options...
maddoctor.2738 Posted September 8, 2020 Share Posted September 8, 2020 @Inculpatus cedo.9234 said:Amazon has a Tier 1 backbone; data travels through this, out of this, and then to the AWS servers.It is odd that some players in the same map (same IP address) experience issues while others do not.One would suppose that all players on a faulty map/server/IP address would all experience the issue. Are you positive it doesn't affect everyone on the same map? Link to comment Share on other sites More sharing options...
Inculpatus cedo.9234 Posted September 8, 2020 Share Posted September 8, 2020 I am positive. Link to comment Share on other sites More sharing options...
maddoctor.2738 Posted September 8, 2020 Share Posted September 8, 2020 @Inculpatus cedo.9234 said:I am positive. How much of an effect latency will have on someone depends heavily on their own computer, connection and perception.Someone that plays with under 30 fps will barely notice a spike in latencySomeone that plays with 300 ping will also barely notice a spike in latencyMeanwhile, someone playing with stable 60 fps and 50 ms ping will notice even a much smaller spike in latency.As an example, using GeForce Now I had a latency of 13 ms, topping to 24 ms. However at certain times it went up to 38 to 40 ms, that WAS a latency spike (doubling my normal latency) but in action I could barely feel the effect on my gameplay.So I'm taking any comments about players not affected by latency spikes with a grain of salt Link to comment Share on other sites More sharing options...
Sylviana.1085 Posted September 8, 2020 Share Posted September 8, 2020 To maybe give some more info:I've had these spikes in Desert Highlands and Jahai Bluffs. (Don't really play that much open world, but found myself there for hp's / requiem collection)Each time I had a friend in voice (in discord).It seemed like the lag spikes happened at the same time for both of us (we were on the same instance of the map).It was a different friend for DH and a different friend for Jahai.So I doubt this is smthing local since the spikes happened at the same time.This has really made me hate doing open world... My friends and I don't have these issues while raiding for example. Link to comment Share on other sites More sharing options...
Lan Deathrider.5910 Posted September 8, 2020 Share Posted September 8, 2020 This may be an issue with your home's internet infrastructure. I HAD good performance, even diving into zergs with performance ramped up no problem and no lag, then sometime earlier this year I began to stutter and slow to a crawl even before the zerg renders. I even lap spike to the point where the screen freezes for 30s while I frantically hit ctr-alt-delete to get a response. Turns out the lines from Comcast going into my home, and the lines in my home are 30 years old and no longer to code. They replaced the lines going from the conduit on the street to my house, but can't do the interior work until post Covid, and until they do that part I can't zerg surf anymore...So, TLDR it may very well be your internet infrastructure is aging and out of date. Call your ISP and see if they can come out and check/replace the lines. Link to comment Share on other sites More sharing options...
maddoctor.2738 Posted September 8, 2020 Share Posted September 8, 2020 @Lan Deathrider.5910 said:So, TLDR it may very well be your internet infrastructure is aging and out of date. Call your ISP and see if they can come out and check/replace the lines.How is the aging infrastructure explaining that only Guild Wars 2 is having issues? Link to comment Share on other sites More sharing options...
Ayrilana.1396 Posted September 8, 2020 Share Posted September 8, 2020 @Zok.4956 said:@"Ayrilana.1396" said:Because that’s not how it works. Servers generally do not pick and choose who has issues and who doesn’t. If there’s a series of input lag issues caused by a server, it for the most part affects everyone connected to it. The fact that some players consistently do not experience any issues when on the exact same map instance doing the exact same meta means that it’s unlikely to be a server issue. There are from time to time software/systems bugs in GW2 that cause that one version of a map is in a buggy state, and other versions (different IP) of the same map are not and only players that are on the buggy map (IP) experience problems. This is true; however, not really the case here as I have been on the same map instance doing the same meta as those claiming to have lag and yet my connection is fine. My skills go off when I press them. The fact that some players do not think they experience problems does not prove anything about the problems that other players experience.It's not about disputing whether players are in fact experiencing problems. It's about the source of said problems.Only players/accounts that are experiencing the problems can be used to make tests and to try to locate the problems a little bit more with testing.That's fairly obvious as you can't test for a problem if there isn't one. However, those that do not experience any problems should not be ignored as their situations could help narrow down the the potential sources of the issue and/or rule out others.And we can not know from the outside (no admin/monitoring access to Anets systems) if it is a "server/instance issue" or a "network issue" or a "software issue". In fact, it could be anything and I am sure that there are players where their own bad Wifi/WLAN is causing the problems. But this also does not prove, that there are not also other problems on Anets side. So, until we can make tests, we can not rule out anything.Be aware that this reasoning of yours can also be used against you saying that it's the servers. In a few posts in one of these threads, I suggested it could be an issue with the maps themselves; not the actual IP instance/servers. I'm still unsure though how certain players can consistently be singled out while others remain completely unaffected. If anything, it's more likely to be a software/programming issue than a hardware one. I do experience often skill lag spikes (some several seconds long) in Thunderhead Peaks. So I made some tests during the time the lag spiked happend in my GW2 client and run a MTR to the map IP is was on. During the time the lag spikes happenend, the whole way the packets travel from my PC to the (first) IP of the Amazon network was clean. I had a small RTT (packet round-trip-time) of a few ms (milliseconds) without any spikes of long RTT and MTR had not one lost packet. During the whole time.I also made tests and saw that it is is somhow map specific. On some maps (i.e. core maps) I normally have no skill lag/lag spike, but in S4 maps (especially in Thunderhead Peaks) I had very often skill lag (spikes) of several seconds.Because we have several PC at home I also tested this with two PCs at the same time (one in core maps, one in Thunderhead Peaks, and I also switched the maps/PCs to make sure it is not PC/Hardware specific)): The PC that was in Thunderhead Peaks had skill lag (spikes), the other in core-maps had not.The MTR to the map IPs of the Thunderhead Peaks map and the core map travelled the same way through the internet to the same Amazon IP of the Amazon gateway.So, I am pretty sure, that, during my test, definitely not my internet-connection/access and definitely not "the Internet" (faulty or overloaded providers or provider interconnections) between my PC and the gateway of the Amazon network (Amazon cloud, or whatever you call it) caused the skill lag spikes. So, what else could it be?The only part the packets travel that I could not test further was inside the "Amazon Cloud/Network/Servers" where Anets game servers are.As a result, logic dictates it is has to be somewhere located in the "Amazon cloud" (network, servers, software) and as a result it is the responsibility of Anet to fix this.I only talk about my own tests. I can not prove/disprove if this is true for other players. I am sure there are players with bad WLAN/Wifi that is causing the same problems.It could still be your home network or the PC depending on your configuration. It could also be a programming issue with the maps themselves. Also be aware that you can have a clean connection when using programs like MTR and Pingplotter but the issue can still be due to the ISP network. This could be due to some form of limiting. ICMP packets are typically 64 bytes but the actual packets for the game can get much larger. I admit this is something that I would need to do more research on.Because that’s not how it works. Servers generally do not pick and choose who has issues and who doesn’t.Well, of course servers do this. There could be a lot of reasons, why only some users have issues/server problems and others don't. And the more complicated the server structure, the more complicated it could be to find the reasons and to fix the issues. The hardest issues to find are the ones that do not happen allways. And sometimes, the servers "pick and choose" in a nondeterministic way. Example: race conditions. https://en.wikipedia.org/wiki/Race_conditionI unfortunately read that entire wiki page and don't see what it has to do with the topic at hand.What is different between PoF/S4 maps and other maps, that could be the reason? The map itself (how complicated it is to process it on server side and client side, how much data flows between client and server, the software code) and of course the IP of the map server (which probably means, that it is handled by a different map server/instance).As I stated above, it could very well be the the programming of the map themselves. I find it unlikely to be the servers. It is not unusal (it has happened often in the past) in GW2, that some maps (map IPs) are stuck in some buggy state (caused by overload, software bugs etc) and other versions of the same map (other IPs) have no problem at all. As a result: Some players experience problems, others do not.True. However, there are players on the exact same map instance do the same meta and experience no issues. Unlike the buggy issues in your map which affect all players, this lag isn't having the same impact.I do not know the internal structure of the Anet game servers to make better educated guesses about the reason why this exact problems happen not to all players. It could be related to time-of-day or how much players are on the same map, or how much map-instances are running at the same time. Or something else. Whatever. But I do know that my own tests ruled out "The Internet" and my own equipment. So it has to be somwhere located in the Amazon network (Amazon/Anet) and it is the responsibility of Anet to fix this.It could be or it could not be. I still need to do more research on whether packets from programs such as MTR can go through just fine but larger packets from games may not. It's something well beyond my knowledge so I may end up needing to post on some technical forums.Edit:Read up some more and programs such as MTR use a network level protocol called ICMP. This is different from transport level protocols such as UDP and TCP. The difference is that ICMP is typically not used to exchange data between systems. Perhaps there’s a way to test using the same protocol that GW2 data goes through? Possibly with similarly sized packets?Edit2:It also looks like the route that packets get sent back to you isn’t necessarily the same route you see when you sent them. They may go an entirely different route which could be the cause of latency. This backs up what I have been reading when it’s best to test on both sides. Link to comment Share on other sites More sharing options...
Lan Deathrider.5910 Posted September 8, 2020 Share Posted September 8, 2020 @maddoctor.2738 said:@Lan Deathrider.5910 said:So, TLDR it may very well be your internet infrastructure is aging and out of date. Call your ISP and see if they can come out and check/replace the lines.How is the aging infrastructure explaining that only Guild Wars 2 is having issues?I didn't imply it was only GW2, I occasionally have buffering issues with video. Both my wife and I have phone plans that will call over WiFi if there is a signal, and we have issues there as well. The Comcast guy estimated that we were only getting half of the speed we are supposed to be getting on average for instance. My point is that if your internet lines are from the first few waves of line that were laid down then that very well could be one of the main issues with performance in regards to latency and ping. Link to comment Share on other sites More sharing options...
Teratus.2859 Posted September 8, 2020 Share Posted September 8, 2020 Last big lag spike I had a few days ago I ran a connection test on my internet at the same time and it was running perfectly.There wasn't even much happening on the map either.. like 6 of us duing Junundu Rising that's it.My Internet has nothing to do with Gw2 running like garbage and I don't think it's anything on Anet's end either.My money's on Amazon's servers, I've found threads going back years where people complain about ping issues since the game was moved to AWS. Link to comment Share on other sites More sharing options...
Ramooh.3286 Posted September 9, 2020 Share Posted September 9, 2020 Game is unplayable so I'm here to complain. It's interesting that my ping (as shown in the options menu) rarely gets above 125. Even if I have 5 second delay on commands it will say my ping is 90. I never have problems when the game is trying to convince me to move to a more populated map. Link to comment Share on other sites More sharing options...
MattDu.7123 Posted September 9, 2020 Share Posted September 9, 2020 post removed Link to comment Share on other sites More sharing options...
Zok.4956 Posted September 9, 2020 Share Posted September 9, 2020 @Ayrilana.1396 said:@Zok.4956 said:@Ayrilana.1396 said:Because that’s not how it works. Servers generally do not pick and choose who has issues and who doesn’t. If there’s a series of input lag issues caused by a server, it for the most part affects everyone connected to it. The fact that some players consistently do not experience any issues when on the exact same map instance doing the exact same meta means that it’s unlikely to be a server issue. There are from time to time software/systems bugs in GW2 that cause that one version of a map is in a buggy state, and other versions (different IP) of the same map are not and only players that are on the buggy map (IP) experience problems. Because that’s not how it works. Servers generally do not pick and choose who has issues and who doesn’t.Well, of course servers do this. There could be a lot of reasons, why only some users have issues/server problems and others don't. And the more complicated the server structure, the more complicated it could be to find the reasons and to fix the issues. The hardest issues to find are the ones that do not happen allways. And sometimes, the servers "pick and choose" in a nondeterministic way. Example: race conditions. https://en.wikipedia.org/wiki/Race_conditionI unfortunately read that entire wiki page and don't see what it has to do with the topic at hand.You wrote, that servers do not "pick and choose", and I gave you one example that they can actually do that, even if their behaviour seems random to the persons involved.And with "servers" I always mean a black-box (a combination of hardware and software). No server hardware runs without software.Be aware that this reasoning of yours can also be used against you saying that it's the servers. In a few posts in one of these threads, I suggested it could be an issue with the maps themselves; not the actual IP instance/servers. I'm still unsure though how certain players can consistently be singled out while others remain completely unaffected. If anything, it's more likely to be a software/programming issue than a hardware one. Well, the "maps themselves" are the (virtual) servers, that handle the specific maps. :)I do not believe that the server-hardware itself is faulty. Of course it is somewhere in software or maybe sometimes the server-capactity (this includes hardware and software capabilities) is reached/exhausted and there is some overload-situation. It surely is more complicated because Anet uses Amazon "servers" and Anet probably has not their own/exclusive hardware anymore.It also looks like the route that packets get sent back to you isn’t necessarily the same route you see when you sent them. They may go an entirely different route which could be the cause of latency. true. But the packets that the "Amazon-cloud-gateway" (for lack of a better wording) is sending me as responses to my UDP/ICMP requests will likely travel the same way through the internet the packets from the GW2 servers travel. But of course, to make sure we would have to need access to Anets servers, or a public "looking glass server" from within the Amazon network. Which I do not have.Perhaps there’s a way to test using the same protocol that GW2 data goes through? Possibly with similarly sized packets?Well, there is a possibility, that the packet-sizes could be related to this. GW2 uses TCP as communication protocol between client and map-server. So I did more tests.I did now make further tests to look at the TCP-packet-flow between the actual game-client and the game-server (the map-server). To make it clear: I did NOT look into the GW2-packets (the TCP payload) because that would probably be a violation of the TOS and I do not want to violate the TOS. I just looked at the traffic pattern of the TCP-flow. Thats allowed.I first wanted to establish a feeling and a baseline of the traffic-flow-pattern between client and server with several maps (home instance, core-maps, Special Forces Training Area). The base pattern is:Result/baseline is: When I am idle (my character is standing still and no skill is used) the game-server still sends the client several packets per second (probably with updates about changes in the game world, moving players/npc etc). Only when I am idle and nothing happens/moves in the world (like in the "Special Forces Training Area" instance when I do nothing) there is a pattern of regular intervals of 3-5 seconds duration, where the server does not send any data. Of course if I move my character or use skills, there are a lot of packets per second, that are transmitted server->client and client->server.I then went to Thunderhead Peaks map and landed directly before the last boss fight of the north meta (so the map was not completely full). I did not experience the large lag spikes of several seconds but it felt like I had a few smaller skill lags. Afterwards I looked at the tcp-packet-flow I recorded with Tcpdump/Wireshark. Because a lot was going on during the meta (a lot of other players were moving, fighting and using their skills, a lot of mobs moving etc...) I expected to see a constant flow (several packets per second) from the server to the client.But actually I found a few locations in the flow where the client did not receive any packets from the server for more than a second each.If the client does not get packets from the server he can not update its own skill-display and the movement of the game-world. etc. Which is one cause for skill-lag.This was not normal behaviour. To have no update for over a second during boss fight is bad. What could be the reason for that? I can think of three possible causes:a) one or more packets from the server to the client were lost somewhere on the way (LAN, ISP, Internet, Amazon, etc...)b) one or more packets from the server to the client were delayed somewhere on the way (LAN, ISP, Internet, Amazon, etc...)c) during the time no packets were send to the client from the server (Anet, map, map-server, Instance,...)For a)I looked into the TCP-flow in detail around the locations with the delay and looked (because TCP handles it, if packets are lost) if there are sequence-gaps, retransmissions etc. (thats simplified, it is complicated to make sure I miss nothing). There is nothing that indicates packet loss of the underlying IP-layer. To make a time-consuming process short: There was NO packet loss involved.For b)If the server sends constantly 4 packets per second (as an example) and all 4 packets are delayed in one second, they would be received in the following second. But in the following second the client would not receive only the 4 packets from the second before, but also the 4 packets the server sends in the next second. So there would be some receive-pattern like:second 1: 4 packetssecond 2: 4 packetssecond 3: 0 packets (here is the packet-delay for a second)second 4: 8 packetssecond 5: 4 packetsBut with all my examinations I could not find a pattern remotely similar to this. So: A packet delay during network transport is NOT the cause.Because I can exclude (a) and (b) it has to be ©.Summary: The map-server (hardware, software, virtual-system, etc.) sometimes does not send data-packets to the game-client, this is not a normal server behaviour on a map where something happens and it results in a client that can not update its own state (skill activations, position of other players/npc etc...) during this communication-silence and this is a reason for skill-lag. In my own actual, last test the gaps were a little bit longer than one second, but I am sure If I would make the same test on a map where I would experience a large skill-lag of several seconds, I would see that the server does not send packets for this several-seconds to the client. This would be consistent and probable.Why is the "server" (hardware+software) doing this (sometimes)? I have no idea. It is Anets job to find the "why" and fix it. Link to comment Share on other sites More sharing options...
Terek.8923 Posted September 9, 2020 Share Posted September 9, 2020 This has made me avoid PoF/LWS4/Icebrood maps completely unless I am playing on my tankiest class Reaper cause I know even with 5 second skill delay my Necro wont die. But its not enjoyable at all. So most of the time I just stick to central Tyria, WvW/SPvP and HoT because I never get skill lag in any of those maps.I see people saying this is a recent development, but I have had this issue since PoF first launched albeit with far lesser skill lag duration and occurred less often. Now its 100% of the time in those maps I get it.PoF has been out for 3 years now and its still not fixed...... I wonder if it ever will :anguished: Link to comment Share on other sites More sharing options...
SdoRy.6839 Posted September 9, 2020 Author Share Posted September 9, 2020 @Zok.4956 said:@Ayrilana.1396 said:@Zok.4956 said:@Ayrilana.1396 said:Because that’s not how it works. Servers generally do not pick and choose who has issues and who doesn’t. If there’s a series of input lag issues caused by a server, it for the most part affects everyone connected to it. The fact that some players consistently do not experience any issues when on the exact same map instance doing the exact same meta means that it’s unlikely to be a server issue. There are from time to time software/systems bugs in GW2 that cause that one version of a map is in a buggy state, and other versions (different IP) of the same map are not and only players that are on the buggy map (IP) experience problems. Because that’s not how it works. Servers generally do not pick and choose who has issues and who doesn’t.Well, of course servers do this. There could be a lot of reasons, why only some users have issues/server problems and others don't. And the more complicated the server structure, the more complicated it could be to find the reasons and to fix the issues. The hardest issues to find are the ones that do not happen allways. And sometimes, the servers "pick and choose" in a nondeterministic way. Example: race conditions. https://en.wikipedia.org/wiki/Race_conditionI unfortunately read that entire wiki page and don't see what it has to do with the topic at hand.You wrote, that servers do not "pick and choose", and I gave you one example that they can actually do that, even if their behaviour seems random to the persons involved.And with "servers" I always mean a black-box (a combination of hardware and software). No server hardware runs without software.Be aware that this reasoning of yours can also be used against you saying that it's the servers. In a few posts in one of these threads, I suggested it could be an issue with the maps themselves; not the actual IP instance/servers. I'm still unsure though how certain players can consistently be singled out while others remain completely unaffected. If anything, it's more likely to be a software/programming issue than a hardware one. Well, the "maps themselves" are the (virtual) servers, that handle the specific maps. :)I do not believe that the server-hardware itself is faulty. Of course it is somewhere in software or maybe sometimes the server-capactity (this includes hardware and software capabilities) is reached/exhausted and there is some overload-situation. It surely is more complicated because Anet uses Amazon "servers" and Anet probably has not their own/exclusive hardware anymore.It also looks like the route that packets get sent back to you isn’t necessarily the same route you see when you sent them. They may go an entirely different route which could be the cause of latency. true. But the packets that the "Amazon-cloud-gateway" (for lack of a better wording) is sending me as responses to my UDP/ICMP requests will likely travel the same way through the internet the packets from the GW2 servers travel. But of course, to make sure we would have to need access to Anets servers, or a public "looking glass server" from within the Amazon network. Which I do not have.Perhaps there’s a way to test using the same protocol that GW2 data goes through? Possibly with similarly sized packets?Well, there is a possibility, that the packet-sizes could be related to this. GW2 uses TCP as communication protocol between client and map-server. So I did more tests.I did now make further tests to look at the TCP-packet-flow between the actual game-client and the game-server (the map-server). To make it clear: I did NOT look into the GW2-packets (the TCP payload) because that would probably be a violation of the TOS and I do not want to violate the TOS. I just looked at the traffic pattern of the TCP-flow. Thats allowed.I first wanted to establish a feeling and a baseline of the traffic-flow-pattern between client and server with several maps (home instance, core-maps, Special Forces Training Area). The base pattern is:Result/baseline is: When I am idle (my character is standing still and no skill is used) the game-server still sends the client several packets per second (probably with updates about changes in the game world, moving players/npc etc). Only when I am idle and nothing happens/moves in the world (like in the "Special Forces Training Area" instance when I do nothing) there is a pattern of regular intervals of 3-5 seconds duration, where the server does not send any data. Of course if I move my character or use skills, there are a lot of packets per second, that are transmitted server->client and client->server.I then went to Thunderhead Peaks map and landed directly before the last boss fight of the north meta (so the map was not completely full). I did not experience the large lag spikes of several seconds but it felt like I had a few smaller skill lags. Afterwards I looked at the tcp-packet-flow I recorded with Tcpdump/Wireshark. Because a lot was going on during the meta (a lot of other players were moving, fighting and using their skills, a lot of mobs moving etc...) I expected to see a constant flow (several packets per second) from the server to the client.But actually I found a few locations in the flow where the client did not receive any packets from the server for more than a second each.If the client does not get packets from the server he can not update its own skill-display and the movement of the game-world. etc. Which is one cause for skill-lag.This was not normal behaviour. To have no update for over a second during boss fight is bad. What could be the reason for that? I can think of three possible causes:a) one or more packets from the server to the client were lost somewhere on the way (LAN, ISP, Internet, Amazon, etc...)b) one or more packets from the server to the client were delayed somewhere on the way (LAN, ISP, Internet, Amazon, etc...)c) during the time no packets were send to the client from the server (Anet, map, map-server, Instance,...)For a)I looked into the TCP-flow in detail around the locations with the delay and looked (because TCP handles it, if packets are lost) if there are sequence-gaps, retransmissions etc. (thats simplified, it is complicated to make sure I miss nothing). There is nothing that indicates packet loss of the underlying IP-layer. To make a time-consuming process short: There was NO packet loss involved.For b)If the server sends constantly 4 packets per second (as an example) and all 4 packets are delayed in one second, they would be received in the following second. But in the following second the client would not receive only the 4 packets from the second before, but also the 4 packets the server sends in the next second. So there would be some receive-pattern like:second 1: 4 packetssecond 2: 4 packetssecond 3: 0 packets (here is the packet-delay for a second)second 4: 8 packetssecond 5: 4 packetsBut with all my examinations I could not find a pattern remotely similar to this. So: A packet delay during network transport is NOT the cause.Because I can exclude (a) and (b) it has to be ©.Summary: The map-server (hardware, software, virtual-system, etc.) sometimes does not send data-packets to the game-client, this is not a normal server behaviour on a map where something happens and it results in a client that can not update its own state (skill activations, position of other players/npc etc...) during this communication-silence and this is a reason for skill-lag. In my own actual, last test the gaps were a little bit longer than one second, but I am sure If I would make the same test on a map where I would experience a large skill-lag of several seconds, I would see that the server does not send packets for this several-seconds to the client. This would be consistent and probable.Why is the "server" (hardware+software) doing this (sometimes)? I have no idea. It is Anets job to find the "why" and fix it.Thank you for your efforts! Link to comment Share on other sites More sharing options...
Illyriel.4952 Posted September 9, 2020 Share Posted September 9, 2020 Adding to the pile of voices in hopes that ArenaNet is going to do something about this. Path of Fire and living season 4 are practically unplayable. Link to comment Share on other sites More sharing options...
TristisOris.2165 Posted September 9, 2020 Share Posted September 9, 2020 @shugawuga.3215 said:and they tell me its a isp provider... any supports always say same thing.. useless to create tickets.ls4 maps works fine for a few days after fix, now same lags as usual. Link to comment Share on other sites More sharing options...
SteepledHat.1345 Posted September 11, 2020 Share Posted September 11, 2020 The issues appears to be, I would assume, related to their virtualized network on AWS. In every traceroute I've run on instance IPs you see a disproportionate delays onces you hit AWS IP ranges. Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now